2- To learn the basic concepts behind the four main functional MRI techniques
for imaging the brain.

3- To learn the current power and limitations of these fMRI techniques.

Introduction
    Magnetic Resonance Imaging (MRI) of the brain is well-recognized for its excellent spatial resolution, allowing neuroanatomic structures to be viewed in sharp detail. Recently, it has become possible to modify a conventional MRI scanner to study the brain’s function as well. This new technology, called functional Magnetic Resonance Imaging (fMRI), has brain researchers excited for several reasons. (Table 1)

First, the most commonly used fMRI technique called BOLD-fMRI  (Blood-Oxygen-Level-Dependent fMRI) potentially offers imaging with a temporal resolution on the order of 100 milliseconds and a spatial resolution of 1-2 millimeters,  which is much greater than that of PET and SPECT scanning ¹. This means that transient cognitive events can potentially be imaged and small structures like the amygdala can be more readily resolved. Next, unlike PET and SPECT, most fMRI techniques are noninvasive and do not involve the injection of radioactive materials so that a person can be imaged repeatedly. This may allow imaging of a patient repeatedly through different disease states (ie. imaging a bipolar patient through manic, depressive, and euthymic states) or developmental changes (ie. learning, cognitive stages of development, stages of grief recovery). It also allows for investigations in healthy children due to the low risk. Third, with fMRI, one can easily make statistical statements in comparing different mental states within an individual in a single session whereas PET and SPECT scans usually rely on making statistical statements about group differences between mental states. Thus, fMRI may be of important use in understanding how a given individual’s brain functions and perhaps, in the future, making psychiatric diagnoses and treatment recommendations. It is, in fact, already starting to being used in presurgical  planning to map vital areas like language, motor function, and memory (see figure 1) ^2-6. Perhaps most important for the future clinical utility of fMRI is that it involves only some upgrading of conventional MRI machines and, thus, may be become widely available.

 
 
    Below, we describe the principles underlying the different types of functional MRI and give examples of how each technology can be used in psychiatry research or clinical practice. The basic principles underlying all types of MRI were discussed in the previous section on structural imaging. Namely, we discuss in this chapter the four main types of functional MRI:

(1) BOLD-fMRI which measures regional differences in oxygenated blood,
(2) perfusion fMRI which measures regional cerebral blood flow,
(3) diffusion-weighted fMRI which measures random movement of water molecules, and
(4) MRI spectroscopy which can measure certain cerebral metabolites noninvasively.

BOLD-fMRI (Blood-Oxygen-Level-Dependent fMRI)
    BOLD-fMRI is currently the most common fMRI technique (Table 2).

Here, the MRI scanner is tuned to resonate and image hydrogen atoms as in conventional MRI; however, T2*-weighted images are performed which take advantage of the fact that deoxygenated hemoglobin is magnetic whereas oxygenated hemoglobin is not ^6-8. Because of the magnetic properties of the unflipped magnetic deoxyhemoglobin molecule which causes rapid dephasing, T2* signal is retained longer in a region when it has more oxygenated blood compared to when there is less oxygenated blood. Thus, an area with more oxygenated blood will show up more intense on T2*-weighted images compared to when there is less oxygenated blood around.
    With this technique, it is assumed that an area is relatively more active when it has more oxygenated blood compared to another point in time ^1,6. This is based on the principle that when a brain region is being used, arterial oxygenated blood will redistribute and increase to this area. This principle has one caveat: there is a time lag of 3-6 seconds between when a brain region is activated and blood flow increases to it ^1,9,10. During this time lag of 3-6 seconds, in fact, the activated areas experience a relative decrease in oxygenated blood as oxygen is extracted by the active regional neurons. Afterward, the amount of blood flowing to the area far outweighs the amount of oxygen that is extracted so that oxygenated blood is now higher. Although images can be acquired every 100 msecs with echoplanar (a type of rapid acquisition) BOLD fMRI, this predictable but time varied delayed onset of the BOLD response limits the immediate temporal resolution to several seconds instead of the 100 msec potential ⁹. In the future, researchers may be able to improve the temporal resolution of fMRI by measuring the initial decrease in oxygenated blood with activation ¹¹.
    BOLD fMRI is a relative technique in that it must compare images taken during one mental state to another to create a meaningful picture. As images are acquired very rapidly (ie. a set of 15 coronal brain slices every 3 seconds is commonly done in our lab), one can acquire enough images to measure the relative differences between two states to perform a statistical analysis within a single individual.  Ideally, these states would differ in only one aspect so that everything is controlled for except the behavior in question ¹⁰. Breiter et al.(1996), for example, scanned Obsessive-Compulsive Disorder patients and healthy controls during activated (ie. holding dirty washcloth) versus rest states (ie. holding clean washcloth) 12 (see figure 2).

 
    BOLD-fMRI paradigms generally have several periods of rest alternating with several periods of activation . Images are then compared over the entire activation to the rest periods (see figure 3). Images obtained over the first 3 to 6 seconds of each period are generally discarded due to the delay in hemodynamic response. Alternating paradigms are used because the signal intensity generated by the MRI scanner drifts with time.
 
    With current technology, fMRI-BOLD is best used for studying processes that can be rapidly turned on and off like language, vision, movement, hearing, and memory ⁹. The study of emotion is hampered by its slow and variable onset and its inability to be quickly reversed ^13,14. Some have, however, succeeded in using this technique to study emotional processes ^12,15. For example, Whalen et al. (1998) ¹⁶ used a backwards masking procedure to present 3 alternating conditions to 10 subjects: (1) a baseline condition where subjects would see a “+” sign, (2) a “happy” (H) condition where subjects would see repeated presentations of 33 msecs of a happy face followed by 167 msecs of a neutral face, and (3) a “fear” (F)  condition where subjects would see repeated presentations of 33 msecs of a fearful face followed by 167 msecs of a neutral face. Here, fearful and happy faces were presented in such a way that 8 of the 10 participants could not  identify them. Despite this unconscious processing, subjects had relatively increased amygdala activation with fearful faces and relatively decreased amygdala activation with the happy faces (see Figure 4).
 

 
    BOLD-fMRI is very sensitive to movement so that tasks are limited to those without head movement, including speaking. BOLD-fMRI is also limited in that artifacts are often present in brain regions that are close to air (i.e. sinuses). Thus, there are some problems in observing important emotional regions at the base of the brain like the orbitofrontal and medial temporal cortices. Another problem is that sometimes observed areas of activation may be located more in large draining veins rather than directly at a capillary bed near the site of neuronal activation ⁶.
    Currently, there are no indications for BOLD-fMRI in clinical psychiatry, although this technique holds considerable promise for unraveling the neuroanatomic basis of psychiatric disease. It may be of potential help in sorting out diagnostic heterogeneity and treatment planning in the future. Neurologists and neurosurgeons are beginning to use this technique clinically to noninvasively map language, motor, and memory function in patients undergoing neurosurgery ^2-6.

Perfusion fMRI
    Two fMRI methods have been developed for measuring cerebral blood flow. The first method, called intravenous bolus tracking, relies on the intravenous (iv) injection of a magnetic compound such as a gadolinium-containing contrast agent and measuring its T2*-weighted signal as it perfuses through the brain over a short time period of time (Table 3) ^17-20 .

 Areas perfused with the magnetic compound show less signal intensity as the compound creates a magnetic inhomogeneity that decreases the T2* signal.  The magnetic compound may be injected once during the control and once during the activation task and relative differences in blood flow between the two states may be determined to develop a perfusion image¹⁷; alternatively, one can measure changes in blood flow over time after a single injection to generate a perfusion map ¹⁹.  Belliveau et al. (1991) used the technique to create the first functional magnetic resonance maps of human task activation using a visual stimulation paradigm. They imaged the occipital lobe after injecting gadolinium-DTPA once during darkness and again during a flashing light to map the visual response. They made a statistical comparison between images obtained during visual stimulation versus those obtained during darkness to generate the now famous image in Figure 5.
 

 

    Although gadolinium-based contrasts are not radioactive, the number of boluses that can be given to an individual is limited by the potential for kidney toxicity with repeated tracer administration. This technique also only generates a map of relative cerebral blood flow, not absolute flow as in the next technique.  Arterial spin labelling is a T1-weighted noninvasive technique where intrinsic hydrogen atoms in arterial water outside of the slice of interest are magnetically tagged (“flipped”) as they course through the blood and are then imaged as they enter the slice of interest (Table 4) ^6,21-24.

Arterial spin-labelling is noninvasive, does not involve an iv bolus injection, and can, thus, be repeatedly performed in individual subjects. Also, absolute regional blood flow  can be measured which cannot be easily measured with SPECT or BOLD-fMRI and requires an arterial line with PET. As absolute information is obtained, cerebral blood flow can be serially measured over separate imaging sessions such as measuring blood flow in bipolar subjects as they course through different disease states ²⁵. Absolute blood flow information may be important in imaging such processes as anxiety which may be hard to turn on and off. For instance, in social phobics, a relaxation task may be imaged on one day and anticipating making a speech may be imaged on the next day. Comparing these separate tasks in different imaging sessions would not be possible with BOLD-fMRI. Figure 6 shows an example of the arterial spin-labelling technique. While there is currently no clinical indication for this technique, it may soon be used clinically to help characterize the different stages of acute ischemic stroke ²⁶.

    At this point, arterial spin-labelling has some limitations in that it takes several minutes to acquire information on a single slice of interest. Therefore, one must have a specific brain region that one is interested in examining. Also, as it currently takes several minutes to acquire a single slice, it  would be tedious obtaining enough images on this slice within a single session to make a statistical statement on a given subject. Thus, this does not appear to be a useful mapping technique within individuals unless scanner acquisition time is shortened.

Diffusion-Weighted Imaging
    Diffusion-weighted imaging is very sensitive to the random movement of 1H in water molecules (brownian movement) ²⁶. The amount of water diffusion for a given pixel can be calculated and is called the apparent diffusion coefficient (ADC). Areas with low ADC values (ie.low diffusion) appear more intense. ADC values are direction sensitive. For instance, if images are taken perpendicular to myelin fiber tracts like the optic chiasm, arcuate fasciculus, or corpus callosum, ADC values will be lower than if the images are taken along the length of these fibers. This is thought to be because there is little diffusion across myelin sheaths ²⁷. Thus, ADC direction sensitivity permits detection of myelination and may allow researchers to understand in greater detail myelin development in infants. On the other hand, this direction sensitivity hampers the study of diffusion in other processes as ADC values differ, depending on the imaging plane (axial, coronal, or sagittal). There are now ways to calculate average ADC values incorporating all planes for each pixel, removing “artifacts” due to the direction of acquisition. Removing the directional diffusion sensitivity has been helpful in studying stroke.
    While it is currently unclear how diffusion-weighted imaging will be useful in studying psychiatric disorders, it hold great promise for changing the clinical management of acute ischemic stroke by potentially refining the criteria for patients most likely to benefit from thrombolytic therapy ( see figure 7) ^28-30.

In the early post-stroke period, ADC values are heterogeneous in the ischemic region and the presence of areas that have only mildly diminished ADC values may indicate salvageable tissue. In this way, diffusion-weighted imaging may help reveal the likelihood of whether thrombolytic therapy may be useful. In addition, while ADC values continue to decrease over the first week post stroke, old strokes have ADC values that are normal or high. This allows distinction of old from new strokes which is often difficult to characterize with structural imaging and clinical exam alone when old and new strokes appear in the same brain region.

MRI Spectroscopy
    MRI spectroscopy (MRS) offers the capability of using MRI to noninvasively study tissue biochemistry (Table 6) ^31,32.

In the conventional and functional MRI techniques listed above, the hydrogen atom in water is the main one that is flipped (resonated). In MRS, either 1H atoms in other molecules or other atoms such as 31P, 23Na, K, 19F, or Li are flipped. Within a given brain region called a voxel, information on these molecules is usually presented as a spectrograph with precession frequency on the x-axis revealing the identity of a compound and intensity on the y-axis which helps quantify the amount of a substance (Figure 8, 1H spectrograph).

 

The quantity of a substance is related to the area under its spectrographic peak; the larger the area, the more of a substance that is present (see Figure 9)
 

The reason why several molecules can be identified and quantified within a single scan is that the resonant magnetic pulse has a bandwidth over a narrow precession frequency range so that it can flip several molecules at once. The signal intensity at each of these precession frequencies can then be identified using a complicated mathematical procedure called a Fourrier transform. For a given precession frequency (or spectrographic peak of a given molecule), information can also be presented spatially as metabolic maps which are created with similar principles to the 1H atom in water spatial map in conventional MRI ³³. The spatial resolution of these maps is generally less than that of conventional MRI as the substance concentration is much less than that of water.  Consequently, the minimum area needed to obtain a visible signal is greater.
    The two most widely used MRS techniques involve either viewing  1H atoms in molecules other than water or 31P-containing molecules ³¹. In 1H MRS, the water signal must first be suppressed as it is much greater than the signal from other 1H-containing compounds and has overlapping spectroscopic peaks with these compounds ¹. Compounds that can be resolved with 1H-MRS  include:

(1) N-acetylaspartate (NAA) which is thought to be a neuronal marker that decreases in processes where neurons die;
(2) lactate which is a product of anaerobic metabolism and may indicate hypoxia;
(3) excitatory neurotransmitters glutamate and aspartate;
(4) the inhibitory neurotransmitter gamma-aminobutyric acid (GABA) ;
(5) cytosolic choline which includes primarily mobile molecules involved in phospholipid membrane metabolism but also small amounts of the neurotransmitter acetylcholine and its precursor choline;
(6) myoinositol which is important in phospholipid metabolism and intracellular second messenger systems; and
(7) creatine molecules  such as creatine and phosphocreatine which usually have relatively constant concentrations throughout the brain and are often used as relative reference molecules (ie. you may see NAA concentration reported as the ratio NAA/creatine in the literature).

Phosphorus (31P) MRS allows the quantification of  ATP metabolism,  intracellular pH, and phospholipid metabolism .  ATP metabolism quantification is possible because ATP is involved in the following reactions :
(1) ATP     ADP + Phosphorus (Pi)
 where the relative concentrations of ATP and Pi can be determined with ³¹ P MRS
and
(2) ADP + phosphocreatine     ATP + Creatine
where the relative concentrations of phosphocreatine and ATP can also be determined with ³¹P MRS .
pH can be measured because H2PO4   H+ + HPO4-2, and the resonance frequency for H2PO4 is different from HPO4-2. As the shift between these two molecules is so rapid, they present as one spectrographic peak.  However, when one of these compounds is present at its equilibrium, the peak is shifted closer to that compound’s true precession frequency, allowing changes in pH to be measured by its position ³⁴. Mobile phospholipids, including phosphomonoesters (PME - putative cell membrane building blocks) and phosphodiesters (PDE - putative cell membrane breakdown products) can also be measured, supplying information on phospholipid membrane metabolism.
    MRS can be used to identify regional biochemical abnormalities. For example, ³¹P-MRS studies of euthymic bipolar patients have revealed decreased frontal lobe PMEs (cell membrane building blocks) compared with healthy controls. However, when bipolar patients become either  manic or depressed, their  PMEs increase. These findings appear to be unrelated to medication treatment (see 31,35 for reviews). The finding of decreased frontal PMEs in euthymic bipolars has also been demonstrated in schizophrenia and speculatively accounts for the finding of decreased frontal lobe metabolism in both of these disorders. The schizophrenia finding also appears to be medication-independent ³².
    MRS may also be of future help in the differential diagnosis of certain psychiatric diseases such as dementia. In normal aging, there is a decrease in PMEs and increase in PDEs ^36,37. In early Alzheimer’s Dementia, there appears to be increased PMEs which may be help distinguish it from normal aging³⁸. Researchers have also found  increased myoinositol and decreased NAA levels in Alzheimer’s Dementia compared with healthy controls ^39,40. Some believe that a decrease in NAA coupled with an increased myoinositol level helps in diferentiating probable Alzheimer’s Dementia from healthy age-matched controls as well as other dementias (usually decreased NAA but normal myoinositol levels) ³⁹.
    With MRS, changes in metabolic activity can be measured over time within an individual scanning session. For instance, Dager et al. (1995) used ¹H MRS to measure changes in lactate concentration with controlled hyperventilation in panic disorder patients and healthy controls ⁴¹. MRS can also be used to measure changes in metabolic activity between sessions, such as before and after medication treatment. For example, Satlin et al. (1997) used ¹H MRS to measure midparietal lobe cytosol choline levels in 12 Alzheimer’s subjects before and after treatment with Xanomeline, an M1 selective cholinergic agonist, or placebo ⁴². Additionally, MRS can be used to measure drug levels of certain psychotropic drugs. The magnetic elements ⁷Li and ¹⁹F do not naturally occur in the human body but they are found in psychotropic drugs; lithium for ⁷Li and fluoxetine and stelazine for ¹⁹F. For example, studies have consistently found that the brain concentrations of lithium are about 0.5 that of serum Li levels and correlate with treatment response.
    For psychiatry, MRS is a research tool at this time. In neurology and neurosurgery, however, MRS is starting to be used in the characterization of tumor, stroke, and epileptogenic tissue and in presurgical planning ^26,34.

Limitations
    MRS is restricted to studying mobile magnetic compounds. As neurochemical receptors are not usually mobile, they cannot be measured with MRS. Thus, receptor-ligand studies are still the domain of SPECT and PET. Another problem with MRS is that due to the low concentrations of many of the imaged substances, larger areas than with water are needed to obtain detectable signals. Larger volume units imaged over longer periods are thus used with this technique, limiting both temporal and spatial resolution compared with conventional MRI, and BOLD-fMRI. However, stronger magnetic fields which can spread out precession frequencies over a wider range may improve this resolution (ie. a magnetic field twice as strong will double the difference between two substances’ precession frequencies, thus increasing resolution). Stronger magnetic fields may also allow detection of compounds that are currently considered to be in too low a concentration to be seen with current MRS equipment.

Conclusions
    While there are currently no clinical indications for ordering any of these fMRI techniques, they hold considerable promise for unraveling the neurocircuitry and metabolic pathways of psychiatric disorders in the immediate future and in helping in psychiatric diagnosis and treatment planning down the road. Their first widespread clinical use will likely be in neurosurgical planning and perhaps, the management of acute stroke. As these techniques are generally noninvasive, can be performed on upgraded conventional MRI scanners, and are less expensive than acquiring a cyclotron to perform PET, they have a greater chance of becoming available in most hospitals over the next several years. Once fMRI techniques are perfected, they will likely offer considerable advantage over PET and SPECT scanning in all aspects except receptor-ligand studies which cannot currently be performed with fMRI. We hope that this article provides you with the necessary background to understand the current fMRI techniques and to more easily evaluate articles in this rapidly developing field.

References

1. Posse S, Muller-Gartner HW, Dager SR. Functional Magnetic Resonance Studies of Brain Activation. Seminars in Clinical Neuropsychiatry 1996; 1:76-88.
 
2. Binder JR, Swanson SJ, Hammeke TA, Morris GL, Mueller WM, Fischer M, et al. Determination of language dominance using functional MRI: A comparison with the Wada test.. Neurol 1996; 46:978-984.
 
3. Binder JR. Neuroanatomy of Language Processing Studies with Functional MRI. Clinical Neuroscience 1997; 4:87-94.
 
4. Bookheimer SY. Functional MRI Applications in Clinical Epilepsy. Neuroimage 1996; 4:S139-S146.
 
5. Jack CR, Thompson RM, Butts RK, Sharbrough FW, Kelly PJ, Hanson DP, et al. Sensory Motor Cortex: Correlation of Presurgical Mapping with Functional MR Imaging and Invasive Cortical Mapping. Radiology 1994; 190:85-92.
 
6. Moseley ME, deCrespigny A, Spielman DM. Magnetic Resonance Imaging of Human Brain Function. Surgical Neurology 1996; 45:385-391.
 
7. Thulborn KR, Waterton JC, Matthews PM, Radda GK. Oxygenation dependence of the transverse relaxation time of water protons in whole blood at high field. Biochem Biophys Acta 1982; 714:265-270.
 
8. Turner R, Le Bihan D, Moonen CT, Despres D, Frank J. Echo-planar time course MRI of cat brain oxygenation changes. Magn Reson Med 1991; 22:159-166.
 
9. David A, Blamire A, Breiter H. Functional Magnetic Resonance Imaging: A new technique with implications for psychology and psychiatry. Brit J Psychiatry 1994; 164:2-7.
 
10. George MS, Ketter TA, Kimbrell TA, Speer AM, Lorberbaum J, Liberatos CC, et al. Neuroimaging Approaches to the Study of Emotion. In: Borod J, editor. The Neuropsychology of Emotion. New York: Oxford University Press, 1998:IN PRESS
 
11. Menon RS, Ogawa S, Hu X, Strupp JP, Anderson P, Ugurbil K. BOLD based functional MRI at 4 Tesla includes a capillary bed contribution: echo-planar imaging correlates with previous optical imaging using intrinsic signals. Magn Res Med 1995; 33:453-459.
 
12. Breiter HC, Rauch SL, Kwong KK, Baker JR, Weisskoff RM, Kennedy DN, et al. Functional Magnetic Resonance Imaging of Symptom Provocation in obsessive-compulsive disorder.. Arch Gen Psychiatry 1996; 53:595-606.
 
13. George MS, Ketter TA, Post RM. What Functional Imaging Studies Have RevealedAbout the Brain Basis of Mood and Emotion. In: Panksepp J, editor. Advances in Biological Psychiatry. Greenwich, Conn. JAI Press, 1996:63-113.
 
14. George MS, Post RM, Ketter TA, Kimbrell TA. Neural Mechanisms of Mood Disorders. In: Rush AJ, editor. Current Review of Mood Disorders. Philadelphia: Current MEdicine, 1995:1
 
15. Whalen PJ, Rauch SL, Etcoff NL, McInerney SC, Lee MB, Jenike MA. Masked presentations of emotional facial expressions modulate amygdala activity without explicit knowledge. The J Neuroscience 1998; 18:411-418.
 
16. Breiter HC, Etcoff NL, Whalen PJ, Kennedy WA, Rauch SL, Buckner RL, et al. Response and habituation of the human amygdala during visual processing of facial expression.. Neuron 1996; 17:875-887.
 
17. Belliveau JW, Kennedy DN, McKinstry RC, Buchbinder BR, Weisskoff RM, Cohen MS, et al. Functional mapping of the human visual cortex by magnetic resonance imaging. Science 1991; 254:716-719.
 
18. Rosen BR, Belliveau JW, Aronen HJ, Kennedy D, Buchbinder BR, Fischman A, et al. Susceptibility contrast imaging of cerebral blood volume: human experience. Magn Reson Med 1991; 22:293-9; discuss.
 
19. Harris GJ, Lewis RF, Satlin A, English CD, Scott TM, Yurgelun-Todd DA, et al. Dynamic susceptibility contrast MRI of regional cerebral blood volume in Alzheimer's disease. Am J Psychiatry 1996; 153:721-724.
 
20. Villringer A, Rosen BR, Belliveau JW, Ackerman JL, Lauffer RB, Buxton RB, et al. Dynamic imaging with lanthimide chelates in normal brain: contrast due to magnetic susceptibility effects. Magn Res Med 1988; 6:164-174.
 
21. Warach S, Sievert B, Darby D, Thangaraj V, Edelman R. EPISTAR perfusion echo-planar imaging of human brain tumors. J MRI 1994; 4:S8
 
22. Edelman RR, Siewert B, Darby DG, Thangaraj V, Nobre AC, Mesulam MM, et al. Qualitative mapping of cerebral blood flow and functional localization with ech-planar MR imaging and signal targeting with alternating radio frequency. Radiology 1994; 192:513-520.
 
23. Roberts DA, Detre JA, Bolinger L, Insko EK, Leigh JS. Quantitative magnetic resonance imaging of human brain perfusion at 1.5 T using steady state inversion of arterial water. Proc Natl Acad Sci USA 1994; 91:33-37.
 
24. Williams DS, Detre JA, Leigh JS, Koretsky AP. Magnetic resonance imaging of perfusion using spin inversion of arterial water. Proc Natl Acad Sci USA 1992; 89:212-216.
 
25. Speer AM, Upadhyaya VH, Bohning DE, Risch SC, Vincent DJ, George MS. New Windows into Bipolar Illness: Serial Perfusion MRI Scanning in Rapid-Cycling Bipolar Patients. APA New Research Abstracts 1997; 111 Abstract.
 
26. Fisher M, Prichard JW, Warach S. New magnetic resonance techniques for acute ischemic stroke. JAMA 1995; 274:908-911.
 
27. Sakuma H, Nomura Y, Takeda K, Tagami T, Nakagawa T, Tamagawa Y, et al. Adult and neonatal human brain: diffusional anisotropy and myelination with diffusion-weighted MR imaging. Radiology 1991; 180:229-233.
 
28. Lutsep HL, Albers GW, deCrespigny A, Kamat GN, Marks MP, Moseley ME. Clinical utility of diffusion-weighted magnetic resonance imaging in the assessment of ischemic stroke. Ann Neurol 1997; 41:574-580.
 
29. Koroshetz WJ, Gonzalez G. Diffusion-weighted MRI: an ECG for "brain attack"? Ann Neurol 1997; 41:565-566.
 
30. Zivan JA. Diffusion-weighted MRI for diagnosis and treatment of ischemic stroke. Ann Neurol 1997; 41:567-568.
 
31. Moore CM, Renshaw PF. Magnetic Resonance Spectroscopy Studies of Affective Disorders. In: Krishnan KRR, Doraiswarmy PM, editors. Brain Imaging in Clinical Psychiatry. New York: Marcel-Dekker, 1997:185-213.
 
32. Frangou S, Williams SC. Magnetic resonance spectroscopy in psychiatry: basic principles and applications. British Medical Bulletin 1996; 52:474-485.
 
33. Posse S, Dager SR, Richards TL, Yuan C, Ogg R, Artru AA, et al. In vivo measurement of regional brain metabolic response to hyperventilation using magnetic resonance: proton echoplanar spectroscopic imaging (PEPSI). Magn Res Med 1997; 37:858-865.
 
34. Duncan JS. Imaging and epilepsy. Brain 1997; 120:339-377.
 
35. Ketter TA, George MS, Kimbrell TA, Benson BA, Post RM. Functional brain imaging in mood and anxiety disorders. Current Review of Mood and Anxiety Disorders 1997; 1:96-112.
 
36. Pettegrew JW, Withers G, Panchalingam K, Post JF. 31P nuclear magnetic resonance (NMR) spectroscopy of brain in aging and Alzheimer's Disease. J Neural Transm Suppl 1987; 24:261-268.
 
37. Pettegrew JW, Panchalingam K, Withers G, McKeag D, Strychor S. Changes in brain energy and phospholipid metabolism during development and aging in the Fischer 344 rat. J Neuropathol Exp Neurol 1990; 49:237-249.
 
38. Pettegrew JW, Panchalingam K, Moossy J, Martinez J, Rao G, Boller F. Correlation of phosphorus-31 magnetic resonance spectroscopy and morphologica findings in Alzheimer's disease. Arch Neurol 1988; 45:1093-1096.
 
39. Miller BL, Moats RA, Shonk T, Ernst T, Woolley S, Ross BD. Alzheimer Disease: depiction of increased cerebral myo-inositol with proton MR spectroscopy. Radiology 1993; 187:433-437.
 
40. Shonk TK, Moats RA, Gifford P, Michaelis T, Mandigo JC, Izumi J, et al. Probable Alzheimer disease: diagnosis with MR spectroscopy. Radiology 1995; 195:65-72.
 
41. Dager SR, Strauss WL, Marro KI, Richards TL, Metzger GD, Artru AA. Proton Magnetic Resonance Spectroscopy Investigation of Hyperventilation in Subjects with Panic Disorder and Comparison Subjects. Am J Psychiatry 1995; 152:666-672.
 
42. Satlin A, Bodick N, Offen WW, Renshaw PF. Brain proton magnetic resonance spectroscopy (1H-MRS) in Alzheimer's Disease: changes after treatment with xanomeline, an M1 selective cholinergic agonist. Am J Psychiatry 1997; 154:1459-1461.
 
43. Cohen BM, Renshaw PF, Stoll AL, Wurtman RJ, Yurgelan-Todd D, Babb SM. Decreased Brain Choline Uptake in Older Adults: An in vivo proton magnetic resonance spectroscopy study. JAMA 1995; 274:902-907.